2.0 KiB
2.0 KiB
name, description, allowed-tools, version
| name | description | allowed-tools | version |
|---|---|---|---|
| validator-expert | Production readiness validator for Vertex AI Agent Engine deployments (ADK agents ONLY). 5-category validation: Security (IAM, VPC-SC, Model Armor), Monitoring (2025 dashboard, alerts, SLOs), Performance (auto-scaling, Code Execution TTL), Compliance (audit logs, data residency), Best Practices (Memory Bank, A2A protocol). Generates weighted production readiness score (0-100%) with PASS/WARNING/FAIL status. Triggers: "validate agent engine deployment", "production readiness", "security audit", "check compliance" | Read, Grep, Glob, Bash | 1.0.1 |
What This Skill Does
Production validator for Vertex AI deployments. Performs comprehensive checks on security, compliance, monitoring, performance, and best practices before approving production deployment.
When This Skill Activates
Triggers: "validate deployment", "production readiness", "security audit vertex ai", "check compliance", "validate adk agent"
Validation Checklist
Security Validation
- ✅ IAM roles follow least privilege
- ✅ VPC Service Controls enabled
- ✅ Encryption at rest configured
- ✅ No hardcoded secrets
- ✅ Service accounts properly configured
- ✅ Model Armor enabled (for ADK)
Monitoring Validation
- ✅ Cloud Monitoring dashboards configured
- ✅ Alerting policies set
- ✅ Token usage tracking enabled
- ✅ Error rate monitoring active
- ✅ Latency SLOs defined
Performance Validation
- ✅ Auto-scaling configured
- ✅ Resource limits appropriate
- ✅ Caching strategy implemented
- ✅ Code Execution sandbox TTL set
- ✅ Memory Bank retention configured
Compliance Validation
- ✅ Audit logging enabled
- ✅ Data residency requirements met
- ✅ Privacy policies implemented
- ✅ Backup/disaster recovery configured
Tool Permissions
Read, Grep, Glob, Bash - Read-only analysis for security
References
- Vertex AI Security: https://cloud.google.com/vertex-ai/docs/security